Effects of gait retraining with real-time biofeedback in patients with knee osteoarthritis: systematic review and meta-analysis

Abstract

Purpose: Use of real time biofeedback during gait retraining has been shown to be an effective form of therapy in stroke patients and other neurological conditions, such as Parkinson’s disease. In recent years, there has been increasing use of biofeedback for gait retraining in orthopedic conditions, with a particular focus on knee osteoarthritis. The aim of our systematic review was to investigate the effects of real-time biofeedback when used during gait retraining in knee osteoarthritis (KOA) patients. The review was focussed specifically on effects on the knee adduction moment (KAM) and patient reported outcome measures, in particular pain and activity limitations. KAM is regarded to be a suitable surrogate measure of the medial knee joint loading, an important parameter in the progression of knee osteoarthritis in this group. Methods: Searches were conducted in MEDLINE, EMBASE, CINAHL, SportDiscus, Web of Science and the Cochrane Register for Controlled Trials from inception to May 2015 for gait plus synonyms and feedback plus synonyms and knee osteoarthritis plus synonyms. Titles and abstracts were screened by one reviewer for studies featuring biofeedback interventions for reducing KAM or improving pain and activity limitations in KOA patients. Studies featuring healthy controls were also included where the feedback under evaluation was being developed with the aim of using it in KOA patients, since we considered that these papers could contain important information about implementation of such feedback programs. Risk of bias was assessed using the Downs and Black checklist by two independent reviewers. Data were extracted from each paper according to predefined criteria. Finally meta-analyses were undertaken for changes in KAM in KOA patients and healthy controls and for changes in pain and function in KOA patients. Results: The electronic search strategy resulted in a total of 190 potentially eligible studies. One additional paper was found through other methods. After screening by title and abstract, 30 full papers were found to be potentially eligible and were retrieved for full review. After assessment by two independent reviewers, 12 papers meeting all of the predefined criteria remained. Except for one randomized controlled trial, all papers included used an uncontrolled within-group design. All studies included in this review reported a reduction in KAM as a result of biofeedback training, although this was not significant at the level of the meta-analysis in the case of KOA patients. Meta-analyses for reduction of KAM in non-pathological subjects revealed moderate to strong evidence for biofeedback in reducing KAM with a within-group weighted mean difference (WMD) of 0.79BWHt (95% CI 0.48–1.09 BWHt), compared to 0.31BWHt (95% CI −0.15–0.77BWHt) for patients with KOA. Regarding reduction in pain and improvement in function (in patients only), the evidence was stronger with standardised mean differences of 0.67 (95% CI 0.28–1.06) and 0.56 (95% CI 0.18–0.95) respectively. Regarding the methods used for implementation of the biofeedback in the studies, most studies used visual feedback although the presentation differed between studies. Feedback on the KAM was used relatively infrequently, with most studies providing feedback on a gait parameter assumed to influence the KAM; in most cases the foot progression angle. Conclusions: The meta-analysis showed a large effect of biofeedback gait training on KAM in healthy controls, while also showing that a similar effect may be more difficult to achieve in KOA patients. Importantly, the meta-analyses provided good evidence to support the use of biofeedback training for reduction in pain and improvement in function in KOA patients. Due to limited between-group comparisons, all meta-analyses were based on within-group comparisons. There were insufficient studies found to draw conclusions about the optimum presentation or implementation of the biofeedback. Despite the limitations in the current studies, large improvements in KAM noted in healthy controls, and improvements in patient reported outcome measures, suggest that there is value in exploring further the use of biofeedback training in patients with KOA. Further studies with larger cohorts (of patients rather than healthy subjects), long term follow-up and full randomization between control and feedback groups are required to corroborate the findings reported here.